Still very much dead and alive: Incoherence in conspiracism as a departure from Bayesian rationality

Abstract

Incoherence has long been considered a hallmark of belief in conspiracy theories, though recent research has cast doubt on this claim. We identify theoretical and methodological shortcomings in previous assessments of incoherence and connect this work to current research on reasoning by modeling conspiracism as a set of probabilistic beliefs and computing incoherence as the departure from Bayesian principles of rational belief. In study 1, reanalyzing data from eight studies involving 8,590 participants, we examined the relationship between conspiracism and belief in mutually incompatible conspiracy theories (e.g. Princess Diana “was murdered” and “faked her own death”). A robust association with conspiracism emerged, with the strongest conspiracy believers showing a 91% probability of being incoherent. In study 2, a preregistered experiment with 469 participants, we investigated whether incoherence is specific to conspiratorial contents related to a person’s worldview, or reflects a more general tendency toward inconsistency. We manipulated the content domain (conspiracy vs. nonconspiracy) and logical relation (contrary vs. contradictory) of incompatible explanations for fictitious vignettes and collected explicit probability ratings. We found evidence for a two-component effect of incoherence: a general component, associated with reduced logical monitoring, that emerges in contradictory pairs, and a content-specific component, associated with worldview-driven endorsement, that emerges in contrary conspiracy pairs. Together, these findings encourage a move from dichotomous judgments of dis/belief to a probabilistic modeling of conspiracy beliefs, grounded in a rigorous definition of logical incompatibility and a Bayesian framework for measuring incoherence.